DE1905680A1 - Signal processing system - Google Patents

Description

Our reference: C 2632

THOMSON-CSP 101, Boulevard Murat, Paris l6e / France

Signal processing system

The invention relates to systems for suitable filtering of frequency-modulated signals by an analog method.

This process called "pulse compression" permits the Realization of radar or sonar systems with a long range, which have a high resolution at great distances exhibit.

Bu / ku

909836/1301

19056B0

It is known for processing electrical frequency modulated Signals dispersive systems, such as electrical circuits, acoustic dispersion lines and optical Diffraction arrangements to use. These bodies are unsuitable if the pulses to be compressed have a duration on the order of seconds and a frequency modulation range of a few hundred Hz. This case occurs especially in underwater acoustics when one tries to realize an underwater sound device (sonar) or an echo sounder with high efficiency. The applied The solution is to subject the received signals to preprocessing that does not change their shape, however, their duration is shortened considerably.

To adapt the properties of the pulses received by a sonar device to the properties of a dispersion system, It is known to use a memory with a magnetic carrier which has two different running speeds having. These speeds are used alternately for recording and reproducing signals. Using a Recording ^ and playback system for preprocessing the signals to be compressed is not without its disadvantage. On the one hand This method can only be used if the ratio of the processing speeds of the information carrier is low, on the other hand is the recording and reproducing device expensive, sensitive and space-consuming.

The invention provides a system with a dispersion arrangement for compressing frequency-modulated signals or for dispersing (delay-dependent processing) of short signals and with an arrangement for converting these signals into signals, multiplicity except for an amplification or weakness

calculation coefficients 909835/1308

chungskoeffizieriten homothetic signals in the time ratio K are ^ where this system is characterized in that the converter arrangement has a digital ring memory for recording a sequence of successive samples or scans of the signals mentioned, a sampling or a sampling and quantization arrangement comprising the aforesaid Samples generated on the basis of the signals, and has a writing arrangement, which the sampling arrangement with the Memory connects, so that in the same the samples are written with a first pole frequency, one on Reading arrangement connected to the memory receives the samples at a second repetition frequency, the ratio of which to first repetition frequency is equal to the homothetic ratio K, a digital-analog converter arrangement the restoration of the signals and whereby a coincidence arrangement couples the reading arrangement with the output of the system, so that an applied signal corresponds to only a single compressed or dispersed signal.

The invention is explained in more detail, for example, with the aid of the figures. Show it

Figure 1 is a simplified circuit diagram of the invention System,

FIG. 2 an explanatory diagram,

Figure 3 is a block diagram of the filter system according to the invention,

Figures k and 5 explanatory diagrams, and

FIG. 6 shows a block diagram of the system according to the invention for Generation of the transmission signal.

90983S / 1308 -

The system according to the invention effects the appropriate filtering of a signal S (- | -) in two separate steps, which successively give rise to signals S (t) and S '(t). In order to clearly distinguish these work steps, beis-pie-Ie for the signals S (t) and S '(t) shown in FIG. It it can be seen that S (t) and S '(t) are characteristic of signals which are present at the input and output of a known dis-

persion system encounters. The signal ^s (~ v ~) ^{is in} Figure 2

not shown, it being assumed that it differs from the signal S (t) only by the choice of a different tent scale. If you divide the variable t by the variable

Before replacing - γ- , where K is a constant, it can be seen that the periods of time are multiplied _{by K, while the frequencies T} zen are divided by K. Therefore, apart from one reinforcement

kungs- or attenuation coefficients the signal S (t) and its homologous signal S (- - ^ - ) in the ratio K homothetic.

In the following the transformation S (t) <p = ± ^s ^ ~ k ~ ^ is called "homothetic" in order to distinguish it from the transformation S (t) f = * S ^f (t), which is called "dispersive" .

From Figure 1, a system can be seen, which the suitable filtering of frequency-modulated electrical signals causes. The system has a dispersion arrangement 1 which converts the signal S (t) into the signal S '(t), as well as a sample transducer arrangement, which the homothetic

cal conversion of the signal to be processed S (- ~ -) into the

XV

Signal S (t) causes. These two conversions make the

duration 909835/1308

■■ -T-r-r

τ duration KT of the signal to be compressed is shortened to - ^ -.

N is an identifier of the input signal. N is equal to the product of the duration and the frequency modulation band of the signal S (- | -) or S (t).

The conversion of the signal ^s (—ip) into ^d & s signal S (t) is effected by means of a ring memory 2, which consists for example of a closed chain of K cells, the content of which is gradually re-stored in a regular sequence. An information element written into cell A circulates in memory 2 according to the arrow and is located at the end of K umspelcherungen in the output cell. Such a memory can be formed by a delay line or by a shift register, its input and output

connected by a loop. The signal S () is from

K of a sample arrangement 3 received, which divides the same into samples E. These samples E. are successively in cell A

of the circulating memory 2 is written by means of a writing arrangement 4 in such a way that the sample E * ₊₁ »which follows the sample E. is written into the cell A when the sample E. is in the

Cell U arrives after the memory has been circulated in the direction of the arrow. A pulse generator 5 controls the arrangements 3 and 4 such that the period θ of the sample output of the

t T *

Signal S (—— ·) differs by the amount - γ- from the time f of the

differentiates the run of the samples in the memory 2; β = χ Χ.

The reading arrangement 6 coupled to the memory 2 distributes the

Samples, which in the cell U at time intervals - ^Js -

follow differently. 909835 / $ 130

follow differently. The output of the reading arrangement 6 emits sequences of K samples which represent parts of the signal ^s (-§--) which correspond to the K last samples emitted by the sampling arrangement J. If one looks at a sequence of K samples at the output of the reading order 6, one sees that their content is partially repeated in the sequences preceding and following the KI. If it is assumed that each sample repeats itself - K times at time intervals' - then the signals emitted by the reading arrangement 6 are redundant and, after processing in the arrangement 1, give rise to a plurality of short pulses, the amplitudes of which increase linearly during the period KT and then decrease linearly during the same time.

In fact, a given signal S (-y-) can only correspond to a series of short pulses, the envelope of which is the autocorrelation function of the signal S (-p-). . . .

In order to achieve this condition, according to the invention, a Coincidence arrangement 7 is provided, which is controlled by generator 5 will. The coincidence arrangement 7 receives the multiple signals S ′ (t), which are available at the output of the arrangement 1 are, and takes samples at time intervals θ, their envelope

generates the autocorrelation function of the signal, S (—s * -).

By reversing the order of the work steps, " the system shown schematically in Figure 1 also for converting a short pulse S '(t) into a long pulse S (--—)

, K suitable. In this case one has to give a short impulse to the arrangement 1, which a first stretched impulse SCt) gives away. The pulse S (t) is then applied to the sampling arrangement

tion 3 is given, which takes samples at time intervals -γ- .

909835/1308 -

The samples obtained are then written into the memory 2 by means of the writing arrangement 4 and are finally received by the reading arrangement 6, which the reading with

a subsequent period Θ '= · % , the arrangement on the coincidence circuit 7 a signal S (- ψ-) with a KN times greater duration than the signal S' (t).

The connections dashed in Figure 1 correspond to the reverse How the system works. The signal to be stretched is applied at I and the stretched signal is available at 0.

FIG. 3 shows a more detailed circuit diagram of a filter system for frequency-modulated electrical signals. So that the description of the operation of this system can be followed more easily, let the signal to be processed be formed, for example, by a frequency-modulated pulse, which has a rectangular envelope with a duration KT equal to one second, a center frequency f of 400 Hz and a modulation range af of 200 Hz , that is, H = KT & f = 200.

The pulse to be compressed is received by a bandpass filter, which the same on a sampler 9 ^over-: wears. A pulse generator IO supplies pulses with a repetition frequency of 1000 Hz to the sampler 9 so that it enirlmmt samples with this frequency. The samples are received by an analog-to-digital converter 11 which generates groups of binary signals corresponding to the algebraic value of the samples. The groups of binary signals occurring at the output of the converter 11 are written into the delay lines 14 with the aid of write arrangements 15, the inputs of which

Respectively 909835/1308

are each connected to the outputs of the converter 11 and the delay line Ik .

As a result of the loops thus formed, the samples circulate in the delay lines 14 with a period V which differs from the writing period θ of the samples by the amount -% -, where K is the number of samples which each of the delay lines 14 can contain.

The number of delay lines connected in parallel is equal to the number of times or steps of the Quantization "arrangement.

At the output of the lines lh , the samples emerge with a repetition frequency which is K times greater than the sampling frequency of the signal to be compressed. The lines 14 feed a digital-to-analog converter 16. The output of the converter 16 emits pulse trains with the duration θ, which contain sections of the signal S (t) homothetic to the signal to be processed, this homothetic signal by a Duration T, a center frequency Kf and a frequency modulation range KAf is characterized. In the example chosen, with K = 1000, one finds T = 1 mS, Kf = 400 kHz and KAf = 200 kHz. The homothetic pulse has the type S (t) shown in FIG. The signals emitted by the converter 16 are suitably filtered by means of a dispersive delay line 19, the line 19 being coupled to the converter via a bandpass filter 17 and a mixer 18 so that the signals are converted into the working range of this line. At the exit of the line 19 grain

90983S / I30I ^prelistened

primed pulses similar to the signal S '(t) in Figure 2. The The duration of these pulses changes as an inverse function of the extent of the modulation range. The minimum value this-

T
this duration is equal to - «- = 5

The pulses emitted by the dispersive line 19 are determined by means of the amplitude detector 20 and are detected given one of the inputs of a coincidence circuit 21. The coincidence circuit 21 is controlled by the generator 10 so that only the signal parts with a duration essentially equal to ΐ

be transmitted. These parts are transmitted at the same time intervals that correspond to the writing period in the Memory 14 correspond. They are put into a filter 22, which summarizes them in the form of a single pulse which is the autocorrelation function of the signal S (-w-) reproduces. In the example given, this function can be expressed by, where B = 200 Hz.

Figures 4 and 5 illustrate the various conversions better recognize which the pulse to be processed is subject to in the course of its progress through the system according to FIG will.

Figure 4 (a) shows part of a sinusoid which has a represents a small section of the signal to be processed. The parallel horizontal lines indicate the quantization levels of the analog-digital converter 11 (here for example l6 levels for four quantization points in time), whereby the vertical lines indicate the times when the sample was taken. In FIG. 4 (b), the outputs from the analog-to-digital converter 11 are shown

Binary information 909835/1308

ORIGINAL INSPECTED

- ίο -

Binary information shown, with each sampling corresponds to a group of four binary digits entering the four delay lines 1 * 1 instead of four prior information items which are erased are written.

FIG. 4 (c) shows the samples written one after the other in the delay lines 1h. These samples correspond to the successive sections of the signal according to Ma) and they repeat with a period X equal to the delay time of the lines 14.

When the binary information contained in the lines 1 * 1 is given to the digital-to-analog converter 16, the latter outputs a sequence of signal excerpts, as shown in FIG. ² Kd). By filtering these sections by means of the filter 17, parts of a sinusoid are restored, as FIG. 4 (e) shows.

FIG. 5 provides an overall overview of the compression method. FIG. 5 (a) shows a frequency-modulated pulse which is divided into 8 sections B, C, D, E, F, G, H, K is.

These sections are subjected to the input of the filter 8 and the output of the mixer 18 to shorten their duration. The shortened sections B ¹ , C ^f , D ', E', F ¹ , G ¹ , H ¹ and K ¹ are shown in Figure 5 (b).

Figure 5 (b) also shows that the shortened sections are repeated in such a way that parts become more and more complete of the pulse to be compressed. If the to

compressing 909835/1308

- li -

When the compressive pulse is completely divided into samples, the following parts become more and more incomplete. By placing the parts b ¹ , B ¹ C, B ¹ C ¹ D ¹ ,... In accordance with FIG. 5 (b) on the dispersive line 19, the sequence shown in FIG of compressed pulses. The height and width of the pulses obtained are related to the frequency modulation range which characterizes each of the parts of the signal shown in Figure 5Cb). By comparing Figures 5 (a) and 5 (c), it can be seen that the initial pulse produced a plurality of highly compressed pulses. In fact, only the tallest and narrowest pulses are of interest since the other pulses correspond to processing incomplete signals. Then only the sequence of the pulses according to FIG. 5 (c) has to be subjected to processing in such a way that it is broken down into samples

Obtains autocorrelation function of S (-γ-). To this end the output of the detector 20 is connected to a circuit 21 which the product of the sequence of Figure 5 (c) with a Sequence of isochronous pulses shown in Figure 5 (d). The product of these two sequences is shown in Figure 5 (e).

The pulses of Figure 5 (d) have a follow-up period equal to ^: Duration of the segments of the signal to be compressed. Its width is equal to that of the narrowest pulse which the detector 20 emits. Since the following period of the pulses shown in Fig. 5 (c) differs from the following period of the pulses shown in Fig. 5 (d), the circuit 21 allows only a small number of narrow pulses with a large amplitude to exist.

the

909835/1308

Figures 3 _S ** and 5 refer to a plant that allows to carry out a compression of a frequency modulated pulse. The pulse to be compressed can be generated by means of known generators, but it is of interest to generate the same on the basis of a short pulse by making the compression arrangement described above operate in the reverse direction. This latter procedure offers the advantage that an excellent correlation is achieved between the signal to be compressed and the devices which carry out the compression.

From Figure 6 lsi; the block diagram of a system can be seen, which enables the expansion of frequency-modulated signals.

The signal to be stretched is a short pulse of the form ■

equal to the pulse S '(t) of Figure 2. An electrical function generator 25 generates a signal which has the center frequency of the dispersive line 26 as the carrier frequency and according to FIG the envelope of the signal S '(t) is amplitude-modulated. At the same time, the function generator 25 emits a synchronization pulse on the synchronizing circuit 24. When the pulse S '(t) is applied to the input of the dispersive line 26, a square-wave signal is obtained at the output of this line the duration T, which is linearly frequency modulated in a band Δ f is. This signal differs from that which the processing in the dispersive line has in mind can be subjected to a frequency modulation, it is therefore necessary to reverse the scale of this signal to be carried out simultaneously with an expansion in the ratio K in order to obtain the signal S (- ^ -).

At the output of the dispersive line 26, the signal is through the

90 98 35/1308 mixer stage

Mixing stage 27 converted in frequency, which the superposition signal of circuit 24 receives. After input into a bandpass filter 28, the stretched and frequency-converted Received pulses from a sampler 29. The sampler 29 gives samples with a specified by the generator 24 Pole period from. These are given to an analog-to-digital converter 30, which the same in combinations of Converts binary signals. Write arrangements 31 receive the binary signals and write them in delay lines 32 with the pole frequency of the sampling.

The lines 32 are provided with a loop which allows the binary signals to circulate. These occur at the exit these lines at the end of a time interval back to j which corresponds to K + 1 times the sampling period. Reading circuits 33, take samples with a sequence Q equal to K times the above sampling period. The phase of The reading times controlled by the circuit 24 allows the samples to be taken in the reverse order with respect to the Write order and therefore the inversion of the signal with respect to the time axis.

The binary signals supplied by the reading circuits 33 are processed by a digital-to-analog converter, which Recovers samples with the same amplitude as those generated by the sampler, but with a K times greater Subsequent period. The signal emitted by the converter 34 is filtered by means of the filter 35. One to the output of the filter 35 connected coincidence circuit 36 cuts a section with the duration KT from the filtered signal. This Excerpt shows the KN times expanded impulse, which in the case of sonar systems is the impulse that is emitted and is received.

For example 909835/1308

For example, a pulse of 1 second can be obtained, which is in a frequency range £ f of 200 Hz a center frequency f of 400 Hz is modulated by assumes the following properties: that of generator 25

The generated pulse P has a duration -rp of 5 / aS and a center frequency of 2 MHz. The dispersive line 26 is adapted to the properties mentioned above. It creates a frequency-modulated square-wave pulse, the duration T of which is equal to 1 ms, which corresponds to a degree of expansion of 200. The stretched pulse is subjected to a frequency change in the mixer 27. The mixer 27, which is excited by a superposition signal of 1.6 MHz, delivers a frequency-modulated pulse between 300 kHz and 500 kHz. The sampling of this pulse available at the output of the filter 28 is carried out at a frequency of 1 MHz. The delay lines 32 have a delay time Έ = 1001 yuS and can each contain 1001 binary signals. The binary signals are extracted with a subsequent period which differs by one microsecond from the delay time 'J', i.e. θ = 1000 jiS. The transmission of the binary signals to the converter 34 takes place regularly at a frequency of 1 kHz. The section cut off by means of the coincidence circuit 36 has a duration of 1 second.

The numerical data given above and the preceding description show that a frequenzmodulier.ten pulse can stretch, then compress, using the same procedure and mostly similar parts. There is an excellent correlation between the signals generated by the system shown in FIG Compression properties of the system shown in FIG. the

combination 909835/1308

The combination of these two systems not only allows a high level of economy to be achieved in terms of number of the parts to be provided, but also ensuring a high degree of correlation due to the reversibility the processing performed on the signals. The alternating Operation as an expansion system when sending and as a compression system when receiving is from the electric Function generator 25 controlled.

The delay lines also fall within the scope of this invention 14 and 32, which form the memory short-circuited with a loop, closed by shift registers substitute. The mode of operation based on the use of binary signals allows high degrees of compression to be achieved. The amplitude range is greater, the more quantization levels are provided in the converters. These levels can vary depending on the nature of the signals to be compressed be equally spaced or not.

The coefficient K of the homothetic transformation can be adjusted by adding a buffer with the matching capability at the input of each digital delay line "P" arranges. By putting the samples in groups of "p" every "q". Inscribes rounds, the coefficient K becomes K '= -9- K.

In addition, the ring memory can be formed by a single delay line in which the groups of binary signals which correspond to each of the samples. With this latter option, the ring buffer is between Arranged shift registers which simultaneously receive the binary signals and can transmit the same one after the other or the other way around. Finally, the one in cascade with the homo-

thetic 909835/1308

The dispersion line used in the thetic compression arrangement can have any characteristic ₃ which relates the group delay time to the frequency. It can be replaced by a suitable electrical circuit or by an optical diffraction arrangement which has the same effect on the signals to be compressed or stretched.

Claims 909835/1308

Claims (1)

Claims IJ System for signal processing which allows the duration of frequency-modulated pulses supplied by a signal source to be changed in two steps, characterized by an arrangement with an input channel and an output channel for converting a signal corresponding to each of these pulses into at least one signal whose duration is in the ratio K is changed and whose waveform remains unchanged, a coicide arrangement with a signal input, which is coupled to the output channel, and with a control output, a control arrangement coupled to the converter arrangement and the control input, wherein each pulse to be processed corresponds to a single one of the signals with changed duration, as well as by a dispersive delay arrangement which is switched into one of the named channels in order to make a different change in the duration of each of these pulses. 2. Plant according to claim 1, characterized in that the converter arrangement comprises: a sampling arrangement with an input for receiving the pulses and an output which outputs a series of samples for each of these pulses, a memory array which receives the samples from each series and supplies a sequence of individual signals, each of which is continuous from a number of successive ones Samples of the series is formed, one of which, called the complete signal, is all samples of the line 909835/1308 Contains series, the dispersive delay arrangement a Input 'for receiving the individual signals and an output for the delivery of the corresponding compressed signal © and the coincidence arrangement that of the complete Signal selects the corresponding compressed signal » 3. Plant according to claim I _3, characterized in that the dispersive delay arrangement is arranged in the input channel, the converter arrangement having: a sampling device with an input for receiving pulses with an extended duration which originate from the dispersive delay arrangement, and with an output for dispensing a series of samples for each of the pulses of extended duration, and a memory device which receives the samples of each row and emits an output signal which lst formed of samples with enlarged sequence intervals _s wherein the coincidence assembly is controlled so that it selects the output signal. H system according to claim 1, characterized in that the converter arrangement a sampling arrangement with an input coupled to the input channel, one with the control arrangement coupled control input and an output as well as a Storage arrangement with one with the output of the sampling drive arrangement coupled input, a control input coupled to the control arrangement and a recoupled to the output channel Has output. 5. Plant according to claim 4, characterized in that the Frobennahmeanordnung has a quantization arrangement which is connected to an analog-to-binary converter arrangement, the output of the sampling arrangement with your Au channel 909835/1308 BATH ORIGINAL channel coupled via a binary-to-analog converter arrangement is. 6. Plant according to claim 4, characterized in that the memory arrangement a write-in arrangement with one with the output, the input coupled to the sampling arrangement, a control input coupled to the control arrangement and an output, a ring memory with one with the output of the write-in arrangement coupled input and an output as well as a reading arrangement with one with the output of the ring memory coupled input, a control input coupled to the control arrangement and an output coupled to the output channel having. \ Installation according to claim 6, characterized in that the ring memory has at least one delay line having an input and an output which are coupled to each other via a feedback loop. & System according to claim 6, characterized in that the registration and reading periods of the memory arrangement differ from one another by a duration equal to the circulation period in the ring memory differentiate. : %. System according to Claim 2, characterized by a first bandpass filter, which is coupled to the input of the sampling arrangement, and by a second bandpass filter, which is arranged between the output of the converter arrangement and the input of the dispersive delay arrangement. 10. Plant according to claim 2, characterized by a frequency change ■ arrangement, which between the output of the converter arrangement tion 909835/1308 tion and the input of the dispersive delay arrangement is arranged. LlT system according to claim 2, characterized by a low-pass • filter, which is coupled to the output of the coincidence arrangement. 12. Plant according to claim 3, characterized by a generator for delivering the pulses at the input of the dispersive delay arrangement. U. system according to claim 3, characterized by a frequency change arrangement, which between the output of the delay arrangement and the input of the sampling assembly is arranged. ft system according to claim 13, characterized by a bandpass filter, which couples the output of the frequency changing arrangement to the input of the sampling arrangement. H ^, system according to claim 3 »characterized by a filter arrangement, which between the Ümsetzeranordriung and the coincidence arrangement is arranged. 909835/130 Ζ4 Leen · it β